Recent advances in video generation can produce realistic, minute-long single-shot videos with scalable diffusion transformers. However, real-world narrative videos require multi-shot scenes with visual and dynamic consistency across shots. In this work, we introduce Long Context Tuning (LCT), a training paradigm that expands the context window of pre-trained single-shot video diffusion models to learn scene-level consistency directly from data. Our method expands full attention mechanisms from individual shots to encompass all shots within a scene, incorporating interleaved 3D position embedding and an asynchronous noise strategy, enabling both joint and auto-regressive shot generation without additional parameters. Models with bidirectional attention after LCT can further be fine-tuned with context-causal attention, facilitating auto-regressive generation with efficient KV-cache. Experiments demonstrate single-shot models after LCT can produce coherent multi-shot scenes and exhibit emerging capabilities, including compositional generation and interactive shot extension, paving the way for more practical visual content creation. Seethis https URLfor more details.
View on arXiv@article{guo2025_2503.10589,
title={ Long Context Tuning for Video Generation },
author={ Yuwei Guo and Ceyuan Yang and Ziyan Yang and Zhibei Ma and Zhijie Lin and Zhenheng Yang and Dahua Lin and Lu Jiang },
journal={arXiv preprint arXiv:2503.10589},
year={ 2025 }
}